CN114549940B

CN114549940B - Image processing method

Info

Publication number: CN114549940B
Application number: CN202210447992.7A
Authority: CN
Inventors: 陈伟璇; 徐海华; 魏溪含; 肖喜中; 杨昭; 赵朋飞
Original assignee: Alibaba China Co Ltd
Current assignee: Alibaba China Co Ltd
Priority date: 2022-04-27
Filing date: 2022-04-27
Publication date: 2022-09-09
Anticipated expiration: 2042-04-27
Also published as: CN114549940A

Abstract

The embodiment of the specification provides an image processing method, wherein the method comprises the steps of obtaining at least one group of unpacking images of each steel scrap briquetting, wherein each steel scrap briquetting is obtained from a target vehicle, and the at least one group of unpacking images of each steel scrap briquetting is obtained from at least three different visual angles; inputting at least one group of unpacking images of each steel scrap pressing block into a detection model to obtain steel scrap proportions and impurity proportions of different grades in at least one group of unpacking images of each steel scrap pressing block; determining the scrap grade and the impurity proportion of each scrap pressing block according to the scrap proportions and the impurity proportions of different grades in at least one group of unpacking images of each scrap pressing block; and determining the scrap grade and the weight deduction and impurity deduction result of the whole vehicle scrap steel pressing blocks in the target vehicle according to the scrap steel grade, the impurity proportion and the weight deduction and impurity deduction model of each scrap steel pressing block.

Description

Image processing method

Technical Field

The embodiment of the specification relates to the technical field of computers, in particular to an image processing method.

Background

The scrap steel is a very important steelmaking raw material, has an important effect on reducing energy consumption and cost, and especially has important significance on energy conservation, emission reduction, production regulation and supply chain safety.

For the waste steel of a certain grade of finished truck pulled into a steel mill, non-steel sundries on the truck are firstly identified, and if the sundries (such as cast steel and the like) which are not allowed to exist, the operation of rejecting and returning goods and the like are required. For other sundries (soil, plastics and the like), estimating the weight as the sundry deduction weight; estimating the weight of the waste steel which is lower than the grade of the waste steel of the whole vehicle and does not reach the standard, and taking the estimated weight as the weight deduction; and then, according to the weight of the scrap steel and the weight of the scrap steel, the scrap steel of the whole vehicle is deducted.

However, at present, the research on grading of scrap steel briquettes and estimation of deduction of the weight of scrap steel is mainly based on scrap steel bulk materials, and scrap steel packing materials, briquettes and the like are not involved; if impurities exist in the middle of the scrap steel packaging material and the briquetting, the identification is difficult, and the adulteration risk is high.

Disclosure of Invention

In view of this, embodiments of the present specification provide an image processing method. One or more embodiments of the present specification also relate to an image processing apparatus, an image processing interaction method, an image processing interaction system, a computing device, a computer-readable storage medium, and a computer program, so as to solve the technical deficiencies of the prior art.

According to a first aspect of embodiments herein, there is provided an image processing method applied to scrap steel briquette processing, including:

acquiring at least one set of unpacking images of each steel scrap pressing block, wherein each steel scrap pressing block is acquired from a target vehicle, and the at least one set of unpacking images of each steel scrap pressing block are acquired from at least three different visual angles;

inputting the at least one group of unpacking images of each steel scrap pressing block into a detection model to obtain steel scrap proportions and impurity proportions of different grades in the at least one group of unpacking images of each steel scrap pressing block, wherein the detection model is a machine learning model;

determining the scrap grade and the impurity proportion of each scrap pressing block according to the scrap proportions and the impurity proportions of different grades in at least one group of unpacking images of each scrap pressing block;

and determining the scrap grade and the weight deduction and impurity deduction result of the whole vehicle scrap steel pressing blocks in the target vehicle according to the scrap steel grade, the impurity proportion and the weight deduction and impurity deduction model of each scrap steel pressing block.

According to a second aspect of embodiments herein, there is provided an image processing apparatus applied to scrap steel briquette processing, including:

an image acquisition module configured to acquire at least one set of unpacking images of each scrap compact, wherein each scrap compact is acquired from a target vehicle and the at least one set of unpacking images of each scrap compact is acquired from at least three different perspectives;

the model processing module is configured to input the at least one group of unpacking images of each steel scrap pressing block into a detection model, and obtain steel scrap proportions and impurity proportions of different grades in the at least one group of unpacking images of each steel scrap pressing block, wherein the detection model is a machine learning model;

the determining module is configured to determine the scrap grade and the impurity proportion of each scrap steel pressing block according to the scrap steel proportion and the impurity proportion of different grades in at least one group of unpacking images of each scrap steel pressing block;

and the result obtaining module is configured to determine the scrap steel grade and the weight deduction and impurity deduction result of the whole vehicle scrap steel pressing blocks in the target vehicle according to the scrap steel grade, the impurity proportion and the weight deduction and impurity deduction model of each scrap steel pressing block.

According to a third aspect of the embodiments of the present specification, there is provided an image processing interaction method applied to an image processing interaction system, the system including at least three photographing devices with different perspectives, an unpacking device, and an image processing apparatus, wherein the method includes:

the image processing device triggers at least three shooting devices with different visual angles under the condition of receiving an unpacking instruction of a user;

the at least three shooting devices with different visual angles acquire panoramic images of each scrap steel pressing block under different visual angles and send the panoramic images to the image processing device, wherein each scrap steel pressing block is acquired from a target vehicle;

the image processing device determines shooting parameters of the at least three shooting devices with different visual angles according to the panoramic image, adjusts the at least three shooting devices with different visual angles according to the shooting parameters, and sends a starting instruction to the unpacking device;

the unpacking device unpacks each scrap steel pressing block according to the starting instruction;

the image processing device receives the adjusted shooting equipment with at least three different visual angles under the condition of receiving the power-off instruction sent by the unpacking equipment for unpacking each scrap steel pressing block, acquires at least one group of unpacking images of each scrap steel pressing block under different visual angles,

inputting at least one group of unpacking images of each steel scrap pressing block into a detection model to obtain steel scrap proportions and impurity proportions of different grades in at least one group of unpacking images of each steel scrap pressing block, wherein the detection model is a machine learning model;

and determining the scrap grade and the weight-deducting and impurity-deducting result of the whole vehicle scrap steel pressing block in the target vehicle according to the scrap grade, the impurity proportion and the weight-deducting and impurity-deducting model of each scrap steel pressing block, and displaying the scrap grade and the weight-deducting and impurity-deducting result of the whole vehicle scrap steel pressing block in the target vehicle to the user through a scrap steel processing interface.

According to a fourth aspect of embodiments herein, there is provided an image processing interactive system comprising at least three photographing apparatuses of different perspectives, an unpacking apparatus, and an image processing apparatus, wherein,

the image processing device is used for triggering at least three shooting devices with different visual angles under the condition of receiving an unpacking instruction of a user;

the shooting equipment at least comprises three pieces of shooting equipment at different visual angles, and is used for acquiring panoramic images of each scrap steel pressing block at different visual angles and sending the panoramic images to the image processing device, wherein each scrap steel pressing block is acquired from a target vehicle;

the image processing device is used for determining shooting parameters of the shooting equipment at the at least three different visual angles according to the panoramic image, adjusting the shooting equipment at the at least three different visual angles according to the shooting parameters, and sending a starting instruction to the unpacking equipment;

the unpacking device is used for unpacking each scrap steel pressing block according to the starting instruction;

the image processing device is used for receiving adjusted shooting equipment with at least three different visual angles under the condition of receiving a power-off instruction sent by the unpacking equipment for unpacking and finishing each scrap steel pressing block, and acquiring at least one group of unpacking images of each scrap steel pressing block at different visual angles,

and determining the scrap grade and the deduction and deduction result of the whole scrap steel pressing blocks in the target vehicle according to the scrap steel grade, the impurity proportion and the deduction and deduction model of each scrap steel pressing block, and displaying the scrap steel grade and the deduction and deduction result of the whole scrap steel pressing blocks in the target vehicle to the user through a scrap steel processing interface.

According to a fifth aspect of embodiments herein, there is provided a computing device comprising:

a memory and a processor;

the memory is configured to store computer-executable instructions and the processor is configured to execute the computer-executable instructions, which when executed by the processor implement the steps of the image processing method described above.

According to a sixth aspect of embodiments herein, there is provided a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the image processing method described above.

According to a seventh aspect of embodiments herein, there is provided a computer program, wherein when the computer program is executed in a computer, the computer is caused to execute the steps of the above-mentioned image processing method.

One embodiment of the present specification implements an image processing method, wherein the method comprises acquiring at least one set of unpacking images of each scrap piece, wherein each scrap piece is acquired from a target vehicle and the at least one set of unpacking images of each scrap piece is acquired from at least three different perspectives; inputting at least one group of unpacking images of each steel scrap pressing block into a detection model to obtain steel scrap proportions and impurity proportions of different grades in at least one group of unpacking images of each steel scrap pressing block, wherein the detection model is a machine learning model; determining the scrap grade and the impurity proportion of each scrap pressing block according to the scrap proportions and the impurity proportions of different grades in at least one group of unpacking images of each scrap pressing block; and determining the scrap grade and the weight deduction and impurity deduction result of the whole vehicle scrap steel pressing blocks in the target vehicle according to the scrap steel grade, the impurity proportion and the weight deduction and impurity deduction model of each scrap steel pressing block.

Specifically, the image processing method adopts a multi-angle imaging technology, acquires unpacking images of steel scrap briquettes from at least three different visual angles, combines a method of integral and local analysis, performs multi-angle analysis on the unpacking process of the steel scrap packing materials and briquettes, automatically identifies the grades of the steel scrap packing materials and briquettes through a detection model, and combines a weight deduction and impurity deduction model to realize automatic calculation of weight deduction and impurity deduction of the steel scrap packing materials and briquettes; meanwhile, the detection and identification of impurities in the unpacking process of the steel scrap packing material and the briquetting can be realized.

Drawings

FIG. 1 is a schematic structural diagram of an image processing method applied to a scrap steel briquette processing scene according to an embodiment of the present disclosure;

FIG. 2 is a flow chart of an image processing method provided in one embodiment of the present description;

fig. 3 is a schematic diagram illustrating an unpacking operation of an unpacking device for a steel scrap briquette in an image processing method according to an embodiment of the present specification;

fig. 4 is a schematic view illustrating a process of identifying foreign matters during unpacking of steel scrap compacts in an image processing method according to an embodiment of the present disclosure;

FIG. 5 is a specific processing diagram of an image processing method applied to a scrap steel briquette processing scenario according to an embodiment of the present disclosure;

FIG. 6 is a schematic view of a scrap steel handling interface in an image processing method according to an embodiment of the present disclosure;

fig. 7 is a schematic diagram illustrating an evaluation result of a target scrap box level deduction model in a scrap box level deduction penalty model updating method of self-iterative learning according to an embodiment of the present specification;

FIG. 8 is a flowchart illustrating a procedure of updating a steel scrap classification penalty model from iterative learning in an image processing method according to an embodiment of the present disclosure;

fig. 9 is a schematic structural diagram of an image processing apparatus according to an embodiment of the present specification;

fig. 10 is a block diagram of a computing device according to an embodiment of the present disclosure.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present description. This description may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein, as those skilled in the art will be able to make and use the present disclosure without departing from the spirit and scope of the present disclosure.

The terminology used in the description of the one or more embodiments is for the purpose of describing the particular embodiments only and is not intended to be limiting of the description of the one or more embodiments. As used in one or more embodiments of the present specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used in one or more embodiments of the present specification refers to and encompasses any and all possible combinations of one or more of the associated listed items.

It should be understood that although the terms first, second, etc. may be used herein in one or more embodiments to describe various information, these information should not be limited by these terms. These terms are only used to distinguish one type of information from another. For example, a first can also be referred to as a second and, similarly, a second can also be referred to as a first without departing from the scope of one or more embodiments of the present description. The word "if" as used herein may be interpreted as "at … …" or "when … …" or "in response to a determination", depending on the context.

First, the noun terms to which one or more embodiments of the present specification relate are explained.

Wagon balance: also known as truck scales, large scales placed on the ground, commonly used to weigh the tons of a truck; the weighing device is a main weighing device for measuring bulk goods of factories, mines, merchants and the like.

Weighing: the truck was driven to a weighbridge for weighing.

Packaging and briquetting of scrap steel: the waste steel which is eliminated or damaged in production and life engineering and is used as recycled waste steel is pressed into rectangular package blocks through multiple compression on a special press machine, so that the transportation and smelting are facilitated.

And (4) weight deduction: metals with substandard scrap cargo levels or rejected items present need to be identified and deducted from the cargo weight.

And (4) impurity buckling: the non-steel materials such as cement, sand and stone exist in the scrap steel cargo, and need to be identified and deducted from the weight of the cargo.

Judging the grade: the scrap steel is classified into different grades according to the thickness, and the system needs to identify the grade of the scrap steel reported by a supplier so as to determine whether the grades are consistent.

Foreign matter: rejected goods (such as plastics and the like) in the scrap goods affect steel smelting and need to be identified.

Judging the grade of a single layer: and judging the grade according to the scrap steel goods on the uppermost layer of the car hopper.

Judging the grade of the whole vehicle: and comprehensively judging the grade of the scrap steel goods of the whole vehicle according to the grade of the scrap steel goods of each layer of the vehicle hopper.

Gross weight: the vehicle itself plus the total weight of the scrap steel compact after the vehicle has been weighed for the first time.

Net weight of scrap steel: and weighing the vehicle for the second time by using the wagon balance, and removing the weight of the scrap steel after the weight of the vehicle is removed.

A settlement system: the enterprise is used for carrying out scrap steel purchase, entry, appraisal and settlement in an integrated information system.

Deployment: refers to the act of publishing algorithms/services to public clouds/hardware terminals.

Scrap steel: the steel scrap material refers to steel scrap materials (such as trimming, end cutting and the like) which are not products in the production process of steel plants and steel materials in used scrapped equipment and components, and the components of the steel scrap materials are steel; the component is pig iron called scrap iron, which is commonly called scrap steel.

The scrap steel is a very important steelmaking raw material, has an important effect on reducing energy consumption and cost, has an important significance in supply side innovation, and especially has an important strategic significance in energy conservation, emission reduction, production regulation and supply chain safety.

At present, the grade of scrap is mainly determined by the thickness of the scrap, and the scrap is divided into a plurality of grades, such as 20mm scrap, 15mm scrap, 10mm scrap, 8mm scrap, 6mm scrap, 4mm scrap, 2mm scrap and the like.

The steel scrap grading process mainly comprises two parts, namely steel scrap grade judgment and steel scrap deduction and impurity deduction, and is a core project of the steel scrap grading system. The weight deduction and impurity deduction of the scrap steel pressing block are mainly to identify and count the unqualified steel products and non-steel products and obtain the weight of the scrap steel needing to be deducted additionally from the scrap steel in a car.

For the whole truck scrap of a certain grade pulled into a steel mill, non-steel sundries on the truck are firstly identified, and if the sundries (such as cast steel and the like) which are not allowed to exist, the operation of refusing and returning the truck is required. For other types of sundries (soil, plastics and the like), estimating the weight as the sundry weight; and estimating the weight of the waste steel which is lower than the grade of the waste steel of the whole vehicle and does not reach the standard, taking the estimated weight as the weight of the deduction, and finally deducting the weight of the waste steel of the whole vehicle according to the weight of the deduction and the deduction.

At present, scrap steel grade judgment, deduction of weight and deduction of impurities and the like are operated by manual and visual estimation and are influenced by various factors such as mood, fatigue state, cognitive difference, interpersonal relationship and the like, manual operation has great instability, the operation process is opaque, the scrap steel grading cost is directly influenced, and loss is caused to a steel mill.

Many enterprises in the market research scrap steel grading technology and weight-deduction and impurity-deduction estimation, but basically mainly scrap steel bulk materials (namely natural forming) are not involved, so that substances in the scrap steel packing materials and the scrap steel briquettes are difficult to identify, namely the scrap steel packing materials and the scrap steel briquettes have hollow or adulteration risks.

In view of this problem, in the present specification, an image processing method is provided. One or more embodiments of the present specification also relate to an image processing apparatus, an image processing interaction method, an image processing interaction system, a computing device, a computer-readable storage medium, and a computer program, which are described in detail one by one in the following embodiments.

Referring to fig. 1, fig. 1 is a schematic structural diagram illustrating an image processing method applied to a scrap steel briquette processing scene according to an embodiment of the present disclosure.

Fig. 1 includes a bale breaker 102, a first camera 104, a second camera 106, a third camera 108, and a server 110 mounted at three views, top, bottom, and top, bottom, of the bale breaker 102.

The first camera 104, the second camera 106, and the third camera 108 may be understood as three types of high-definition cameras with different viewing angles, and the number of the first camera 104, the second camera 106, and the third camera 108 is not limited, and may be set according to actual requirements, for example, the first camera 104 may be set to 2, the second camera 106 may be set to 2, the third camera 108 may be set to 2, and the like.

For convenience of understanding, in the embodiments of the present specification, the numbers of the first camera 104, the second camera 106, and the third camera 108 are each described as one.

Specifically, the vehicle loaded with the scrap steel briquettes enters an unloading area, each scrap steel briquette is placed on a chassis of the bale breaker 102, and a camera is respectively installed at the upper, middle and lower three visual angles of the bale breaker 102: the first camera 104, the second camera 106, and the third camera 108 form a multi-angle shooting sample.

The judgment of a scrap steel briquette and the deduction of a penalty are taken as examples for introduction.

In specific implementation, manual card swiping triggers a unpacking process for the scrap steel pressing blocks, cameras with three visual angles respectively shoot panoramic images of one scrap steel pressing block to obtain panoramic images of the three visual angles of the scrap steel pressing block, the panoramic images of the three visual angles of the scrap steel pressing block are sent to the server 110, and the server 110 adjusts shooting parameters of the first camera 104 and the second camera 106 according to the panoramic images of the three visual angles of the scrap steel pressing block; the unpacking machine 102 unpacks the scrap steel pressing blocks, at the moment that the unpacking machine 102 unpacks and cuts off the power to the scrap steel pressing blocks, the first camera 104, the second camera 106 and the third camera 108 are triggered to shoot unpacking images of the unpacked scrap steel pressing blocks, and after the unpacking machine 102 is electrified for n times, the first camera 104, the second camera 106 and the third camera 108 can shoot n groups of series unpacking images.

The n sets of series unpacking images of the scrap steel pressing blocks are sent to the server 110, and the server 110 can calculate the grade of the scrap steel pressing blocks and the weight of the deduction, the deduction and the addition of impurities according to the n sets of series unpacking images of the scrap steel pressing blocks.

For the grade of the scrap steel pressing blocks and the weight of the fastening weight and the weight of the scrap steel pressing blocks of the whole vehicle, the following embodiment will be described in detail.

The image processing method provided by the embodiment of the specification is applied to a deduction and deduction calculation scene of the scrap steel pressing block, multi-view acquisition of image data can be performed in the scrap steel self-discharging operation through the first shooting device 102, the second shooting device 104 and the third shooting device 106 which are arranged at different positions, and subsequently, the grade judgment of the scrap steel pressing block, automatic deduction and deduction estimation and the like can be quickly and accurately realized through image intelligent analysis based on the abundant images at different angles; the fixed-level work efficiency of steel scrap briquetting is greatly improved, and the use experience of users is improved.

Referring to fig. 2, fig. 2 is a flowchart illustrating an image processing method according to an embodiment of the present disclosure, the image processing method is applied to scrap steel briquetting processing, and specifically includes the following steps.

Step 202: and acquiring at least one group of unpacking images of each scrap steel pressing block.

Wherein each scrap piece is captured from a target vehicle and at least one set of unpacking images of each scrap piece is captured from at least three different perspectives.

The steel scrap pressing block can be understood as a steel scrap pressing block which is formed into a fixed shape (such as a rectangle or a square) after a plurality of pieces of steel scrap are extruded; the target vehicle may be understood as a truck, lorry or the like loaded with a plurality of scrap steel compacts.

In practical application, a plurality of scrap steel pressing blocks exist in a target vehicle, in order to accurately judge the grade and deduct the weight and deduct the impurities of each scrap steel pressing block, the unpacking images of each scrap steel pressing block at different visual angles can be obtained through at least three shooting devices at different visual angles, so that the grade and deduct the weight and deduct the impurities of each scrap steel pressing block can be subsequently judged according to the unpacking images of each scrap steel pressing block. The specific implementation mode is as follows:

the method for acquiring at least one group of unpacking images of each scrap steel pressing block comprises the following steps:

acquiring panoramic images of each scrap steel pressing block at different visual angles through at least three shooting devices at different visual angles;

determining the unpacking position of each scrap steel pressing block according to the panoramic image of each scrap steel pressing block at different viewing angles;

adjusting the shooting parameters of the shooting equipment with at least three different visual angles according to the unpacking position of each scrap steel pressing block;

and acquiring at least one group of unpacking images of each scrap steel pressing block at different visual angles through the adjusted shooting equipment at least three different visual angles.

The shooting device can be understood as any type of shooting device with any resolution, such as a high-definition camera; in the case that the capturing device is understood as a high-definition camera, the capturing device with at least three different viewing angles may be understood as a high-definition camera with at least three different viewing angles, for example, a high-definition camera with three different viewing angles, such as a high-definition camera with three different viewing angles.

Taking a scrap steel pressing block a as an example, at least one group of unpacking images for obtaining the scrap steel pressing block a is described in detail.

Specifically, panoramic images of the scrap steel pressing block a under at least three visual angles, such as panoramic images under three visual angles of Left (Left visual angle), center (center visual angle) and right (right visual angle), are acquired through shooting equipment with at least three different visual angles; calculating the unpacking position of the scrap steel pressing block a according to panoramic images of the scrap steel pressing block a under Left, center and right visual angles; according to the unpacking position of the scrap steel pressing block a, adjusting Left visual angle, center visual angle and right visual angle shooting parameters of shooting equipment (for example, adjusting Left-middle-right three visual angles shooting parameters or adjusting Left-right two visual angles shooting parameters and the like), and finally shooting and acquiring at least one group of unpacking images of the scrap steel pressing block a under the Left visual angle, center visual angle and right visual angle through the parameter adjusted Left visual angle, center visual angle and/or right visual angle shooting equipment.

In practical application, the unpacking operation of one steel scrap pressing block cannot be realized at one time, namely, the unpacking operation of one steel scrap pressing block can not be realized at one time, so that one steel scrap pressing block can be unpacked for many times, and each unpacking operation can obtain a group of unpacking images through at least three shooting devices with different visual angles.

In specific implementation, because the scrap steel pressing block is manually or mechanically placed on unpacking equipment (such as a unpacking machine) in an unfixed way, if the unpacking image is directly shot according to at least three shooting equipment with different visual angles, the unpacking image may not be real due to improper focusing; it is also possible that the unpacked images are distorted due to the fact that the shooting times are not reasonable, and the unpacked images are large or small; therefore, in order to avoid the situation, before the unpacking image of each steel scrap pressing block is acquired, the shooting parameters of the shooting equipment are adjusted according to the unpacking position of the steel scrap pressing block in the unpacking machine at present, so that the accuracy of the unpacking image of the steel scrap pressing block shot by the shooting equipment in the following process is ensured. The specific implementation mode is as follows:

adjusting the shooting parameters of the shooting equipment at least with three different visual angles according to the unpacking position of each scrap steel pressing block, and the method comprises the following steps:

determining a shooting focus position and a shooting magnification according to the unpacking position of each scrap steel pressing block;

and adjusting the shooting parameters of the shooting equipment with at least three different visual angles according to the shooting focus position and the shooting magnification.

The unpacking position of each steel scrap pressing block can be understood as the initial placing position of each steel scrap pressing block on the unpacking device, which is calculated according to panoramic images of each steel scrap pressing block at different visual angles, which are obtained by at least three shooting devices at different visual angles.

According to the previous example, taking the scrap steel pressing block a as an example, firstly, determining the shooting focus position and the shooting magnification according to the unpacking position of the scrap steel pressing block a; then according to the shooting focus position and the shooting magnification, adjusting the shooting parameters of Left (Left visual angle), center (center visual angle) and/or right (right visual angle) shooting equipment; and subsequently, the unpacking image of the scrap steel pressing block a can be more accurately acquired according to the shooting equipment with the shooting parameters adjusted.

In practical application, in order to avoid resource waste caused by the fact that the shooting equipment is always started, the shooting equipment is started again under the condition that an unpacking instruction of a user is received, panoramic images of all the scrap steel pressing blocks at different viewing angles are obtained, and the shooting equipment can be temporarily closed under the condition that the unpacking instruction of the user is not received, so that shooting resources are saved, and the space occupation of the shooting equipment is saved. The specific implementation mode is as follows:

through the shooting equipment at least three different visual angles, acquire the panoramic picture of every steel scrap briquetting under different visual angles, include:

and under the condition of receiving a unpacking instruction of a user, acquiring panoramic images of each scrap steel pressing block at different visual angles through at least three shooting devices at different visual angles.

The unpacking instruction of the user can be triggered by the user through card swiping or clicking operation on a display interface of the unpacking and grading system.

Taking the unpacking instruction of the user as an example that the user triggers through the card swiping operation, under the condition of receiving the unpacking instruction of the user, acquiring panoramic images of each scrap steel pressing block at different visual angles through at least three shooting devices at different visual angles, wherein the condition that the unpacking instruction of each scrap steel pressing block triggered through the card swiping operation of the user is received can be understood as that the shooting devices at least three different visual angles are started, and then the panoramic images of each scrap steel pressing block at different visual angles are acquired through the shooting devices at least three different visual angles.

In order to ensure the accuracy of the grade judgment and the deduction and deduction estimation of each steel scrap pressing block, calculation can be carried out according to the unpacking image of each steel scrap pressing block, so that in order to avoid the situation that the steel scrap pressing block is not unpacked, shooting operation of the steel scrap pressing block is obtained through at least three shooting devices with different visual angles, the steel scrap pressing block image without an actual reference value is obtained, the calculation effect and the calculation efficiency of the grade judgment and the deduction and deduction estimation of each steel scrap pressing block are influenced, and at least three shooting devices with different visual angles can be triggered to shoot the unpacking image of each steel scrap pressing block according to an unpacking and power-off instruction of the unpacking device. The specific implementation mode is as follows:

through the shooting equipment at least three different visual angles after the adjustment, acquire at least a set of bale breaking image of every steel scrap briquetting under different visual angles, include:

and under the condition that power-off instructions sent by unpacking equipment for unpacking each steel scrap pressing block are received, acquiring at least one group of unpacking images of each steel scrap pressing block at different visual angles through the adjusted at least three shooting equipment at different visual angles.

And the power-off instruction is sent out after the unpacking device unpacks the scrap steel pressing block every time.

Still continuing with the above example, in the case of receiving the power-off command sent by the unpacking device after the unpacking of the scrap steel briquette a is finished, the group of unpacking images of the scrap steel briquette a at the three different viewing angles is obtained by shooting the Left (Left viewing angle), the center (center viewing angle) and the right (right viewing angle) shooting devices after the parameters are adjusted. And unpacking equipment unpacks the scrap steel pressing block a for many times, so that the unpacking images of the groups of the scrap steel pressing blocks a can be obtained.

Referring to fig. 3, fig. 3 is a schematic diagram illustrating an unpacking operation of an unpacking device for a steel scrap briquette in an image processing method provided by an embodiment of the present specification.

Fig. 3 includes three parts a, b and c, wherein the clamp shape in the three parts a, b and c is a disassembly claw of the unpacking device, the shooting devices are respectively arranged at the left, middle and right positions of the unpacking device, the part a is a schematic diagram that the scrap steel pressing block is not unpacked by the unpacking device, the part b is a schematic diagram that the disassembly claw of the unpacking device is placed on the scrap steel pressing block and the scrap steel pressing block is to be unpacked, and the part c is a schematic diagram that the disassembly claw of the unpacking device grabs the scrap steel pressing block to be pulled to the left and right, and the scrap steel pressing block is scattered; and the triangular symbol in the middle of the part c is a schematic diagram for finding foreign matters and giving an alarm in the process of unpacking the scrap steel pressing blocks.

Step 204: and inputting the at least one group of unpacking images of each steel scrap pressing block into a detection model to obtain the steel scrap proportion and the impurity proportion of different grades in the at least one group of unpacking images of each steel scrap pressing block.

Wherein the detection model is a machine learning model.

Specifically, after at least one group of unpacking images of each steel scrap pressing block is obtained, steel scrap regions and impurity regions of different grades in each group of unpacking images of each steel scrap pressing block can be obtained through a pre-trained detection model. The specific implementation mode is as follows:

the step of inputting the at least one group of unpacking images of each steel scrap pressing block into a detection model to obtain the steel scrap proportion and the impurity proportion of different grades in the at least one group of unpacking images of each steel scrap pressing block comprises the following steps:

inputting at least one group of unpacking images of each scrap steel pressing block and each unpacking image of each group of unpacking images into a detection model to obtain scrap steel areas and impurity areas of different grades in each unpacking image;

determining the steel scrap proportion and the impurity proportion of different grades in each unpacking image according to the steel scrap regions and the impurity regions of different grades in each unpacking image;

and determining the steel scrap proportion and the impurity proportion of different grades in at least one group of unpacking images and each group of unpacking images of each steel scrap pressing block according to the steel scrap proportion and the impurity proportion of different grades in each unpacking image.

The impurity region is understood to mean a region other than the scrap region and the foreign matter; the Detection model comprises Mask-RCNN (Mask-Regions with CNN features, regional convolutional neural networks), an image instance segmentation algorithm model based on CNN, such as SSD (Single Shot MultiBox Detector), or YOLO (You Only Look On: Unifield, Real-Time Object Detection), and a classified deep neural network algorithm model, such as rest 50 (residual error network), den (classified network).

In specific implementation, inputting each unpacking image of each unpacking image into a detection model in at least one group of unpacking images of each scrap steel briquetting to obtain different grades of scrap steel areas and impurity areas in each unpacking image; determining the steel scrap proportion and the impurity proportion of each unpacking image in different grades according to the steel scrap regions and the impurity regions of each unpacking image in different grades; and calculating the steel scrap proportion and the impurity proportion of different grades in each group of unpacking images in a weighting and averaging mode according to the steel scrap proportion and the impurity proportion of different grades in each group of unpacking images.

Still continuing the above example, taking two sets of unpacking images of the scrap steel briquette a by using the scrap steel briquette a and using cameras with three visual angles of left, middle and right, wherein the first set of unpacking images comprises [ a1, a11 and a111 ]; the second set of images includes [ a2, a22, a222] for example.

The steel scrap areas and the impurity areas of different grades in the two groups of unpacking images of the steel scrap pressing block a are obtained in the mode.

Specifically, three unpacking images [ a1, a11 and a111] in the first unpacking image group are respectively input into a detection model, and a scrap steel region of scrap steel of each grade in the unpacking image a1 and an impurity region of impurities in the unpacking image a1 are obtained; scrap regions of each grade of scrap in the unpacking image a11, impurity regions of impurities in the unpacking image a 11; scrap regions of scrap steel of each grade in the unpacking image a111, impurity regions of impurities in the unpacking image a 111.

Determining the scrap proportion and impurity proportion of each grade of scrap steel in each unpacking image [ a1, a11 and a111] according to the scrap steel regions and impurity regions of the unpacking images [ a1, a11 and a111 ]; the steel scrap proportion of each grade in the unpacking images a1, a11 and a111 is weighted and averaged to obtain the steel scrap proportion of different grades in the first group of unpacking images, for example, the steel scrap proportion of a first grade is 25 percent, and the steel scrap proportion of a second grade is 70 percent; meanwhile, the impurity proportions of the impurities in the unpacked images a1, a11, and a111 are weighted and averaged, so that the impurity proportion of the impurities in the first unpacked image group can be obtained, for example, the impurity proportion is 5%.

Similarly, the scrap steel proportion and the impurity proportion of the scrap steel pressing block of each grade in the second group of unpacking images are obtained by the method.

Step 206: and determining the scrap grade and the impurity proportion of each scrap steel pressing block according to the scrap steel proportion and the impurity proportion of different grades in at least one group of unpacking images of each scrap steel pressing block.

After the steel scrap proportion and the impurity proportion of different grades in each group of unpacking images of each steel scrap pressing block are obtained, the steel scrap grade and the impurity proportion of the steel scrap pressing block can be rapidly calculated according to the steel scrap proportion and the impurity proportion of different grades in each group of unpacking images of the steel scrap pressing block. The specific implementation mode is as follows:

determining the scrap grade and the impurity proportion of each scrap steel pressing block according to the scrap steel proportion and the impurity proportion of different grades in at least one group of unpacking images of each scrap steel pressing block, wherein the steps comprise:

obtaining the scrap grade of each scrap steel pressing block through a linear attenuation weighting mode and an area size voting estimation mode according to the scrap steel proportion of different grades in at least one group of unpacking images and each group of unpacking images of each scrap steel pressing block;

and obtaining the impurity proportion of each scrap steel pressing block in a weighted averaging mode according to the impurity proportion of each group of unpacking images.

Still continuing with the above example, after calculating the scrap proportion and the impurity proportion of each of the two sets of unpacking images of the scrap steel briquette a at different grades, the scrap proportion of each of the scrap steel briquette a at different grades can be calculated in a linear attenuation weighting manner, for example, the scrap proportion of the first grade in the scrap steel briquette a is 25%, and the scrap proportion of the second grade is 70%; and selecting a second grade with 70 percent of scrap steel comparison from the scrap steel proportions of different grades of the scrap steel briquetting a as the scrap steel grade of the scrap steel briquetting a in an area size voting estimation mode. Likewise, scrap grades for other scrap briquettes can be obtained in this way.

And the impurity proportion of the scrap steel pressing block a is determined by the impurity proportion of the two groups of unpacking images of the scrap steel pressing block in a weighting and averaging mode.

In practical application, if unpacking is performed on a certain scrap steel pressing block for n times, n groups of unpacking images are generated; then, for the grade judgment of the scrap steel briquette, firstly, a linear attenuation weighting mode is adopted for fusion, and each group of unpacking images is assumed to contain A, B, C, D four types of scrap steel grades, wherein the scrap steel proportion of each type of scrap steel grade is respectively as follows: RA, RB, RC, RD, the weight calculation formula of each layer in the first group to the nth group is Ri = (n-i)/n; the scrap grade of the scrap steel briquetting is calculated by the following formula: max (sum (RA × i), sum (RB × i), sum (RC × i), sum (RD × i)) (i =1 to n).

Carrying out weighted average on the impurity proportion obtained by dividing the unpacking process for n times to obtain the impurity proportion of the scrap steel pressing block; meanwhile, for the foreign matters (such as cement, plastics, soil and other non-steel products) identified in the unpacking process, an alarm can be given to remind or trigger, and the types and positions of the foreign matters in the scrap steel pressing block can be displayed.

Referring to fig. 4, fig. 4 is a schematic view illustrating a process of identifying foreign matters in a process of unpacking scrap steel compacts in an image processing method according to an embodiment of the present disclosure.

Shown in fig. 4 is that after a certain steel scrap briquette is unpacked, the detection model identifies that foreign matter exists in the steel scrap briquette: and the cement block is used for sending out an alarm prompt.

The user can select the treatment of returning goods of the whole vehicle, partial returning goods, deducting weight and punishing, not paying by the whole vehicle and the like according to the alarm prompt.

Step 208: and determining the scrap grade and the weight deduction and impurity deduction result of the whole vehicle scrap steel pressing blocks in the target vehicle according to the scrap steel grade, the impurity proportion and the weight deduction and impurity deduction model of each scrap steel pressing block.

Specifically, after the scrap grade and the impurity proportion of each scrap steel pressing block are determined, the scrap grade and the weight-deducting and impurity-deducting result of the scrap steel pressing block of the whole vehicle in the target vehicle can be determined according to the scrap grade, the impurity proportion and the weight-deducting and impurity-deducting model of each scrap steel pressing block. The specific implementation mode is as follows:

according to the scrap steel grade, the impurity proportion and the weight deduction and impurity deduction model of each scrap steel pressing block, the scrap steel grade and the weight deduction and impurity deduction result of the whole scrap steel pressing block in the target vehicle are determined, and the method comprises the following steps:

determining the whole vehicle scrap grade and the whole vehicle impurity proportion of the whole vehicle scrap pressing blocks in the target vehicle according to the scrap grade and the impurity proportion of each scrap pressing block;

and determining a deduction and deduction result of the whole vehicle scrap steel pressing block in the target vehicle according to the whole vehicle scrap steel grade, the whole vehicle impurity proportion and the deduction and deduction model.

Specifically, determining the whole vehicle scrap grade and the whole vehicle impurity proportion of the whole vehicle scrap pressing blocks in the target vehicle according to the scrap grade and the impurity proportion of each scrap pressing block comprises:

determining the whole vehicle scrap grade of the whole vehicle scrap pressing block in the target vehicle in an area size voting estimation mode according to the scrap grade of each scrap pressing block;

and determining the whole vehicle impurity ratio of the whole vehicle scrap steel pressing blocks in the target vehicle in an averaging mode according to the impurity ratio of each scrap steel pressing block.

In specific implementation, if a plurality of steel scrap pressing blocks exist in the target vehicle, the steel scrap grades of a preset number of steel scrap pressing blocks can be extracted, and then the steel scrap grade with a larger area is selected as the whole vehicle steel scrap grade of the whole vehicle steel scrap pressing blocks in an area size voting estimation mode; the preset number can be set according to practical application, for example, the preset number can be set to 15 blocks, 20 blocks, and the like; of course, in order to more accurately determine the entire vehicle scrap grade of the entire vehicle scrap briquette, the entire vehicle scrap grade of the entire vehicle scrap briquette can be obtained by voting and estimating the scrap grade of all the scrap briquettes in the target vehicle by area size.

In practical application, for the calculation of the whole vehicle impurity proportion of the whole vehicle scrap steel pressed blocks, a preset number (for example, 15 blocks or 20 blocks) of target scrap steel pressed blocks can be selected from the whole vehicle scrap steel pressed blocks, and the average impurity proportion is calculated to be used as the whole vehicle impurity proportion of the whole vehicle scrap steel pressed blocks.

Then, after the grade of the whole vehicle scrap steel and the proportion of the whole vehicle impurities are determined, the deduction and deduction result of the whole vehicle scrap steel pressing block in the target vehicle can be accurately calculated according to the grade of the whole vehicle scrap steel, the proportion of the whole vehicle impurities and the deduction and deduction model. The specific implementation mode is as follows:

according to the whole vehicle scrap steel grade, the whole vehicle impurity proportion and the deduction and impurity deduction model, the deduction and impurity deduction result of the whole vehicle scrap steel pressing block in the target vehicle is determined, and the method comprises the following steps:

determining the area ratio of the unqualified scrap steel which does not meet the grade of the whole vehicle scrap steel in the whole vehicle scrap steel pressing block in the target vehicle according to the grade of the whole vehicle scrap steel;

determining the mixing degree of the scrap steel briquettes in the target vehicle according to the distribution of the scrap steel grades of each scrap steel briquette in the target vehicle and the impurity proportion of the whole vehicle;

determining the weight of a target vehicle containing a whole vehicle scrap steel pressing block and the weight of a target vehicle not containing the whole vehicle scrap steel pressing block, and determining the difference value of the weight of the target vehicle containing the whole vehicle scrap steel pressing block and the weight of the target vehicle not containing the whole vehicle scrap steel pressing block as the weight of the target vehicle;

and inputting at least two of the substandard area ratio of the scrap steel, the mixing degree of the scrap steel pressing blocks in the target vehicle and the weight of the target vehicle into a deduction and deduction model to obtain a deduction and deduction result of the scrap steel pressing blocks of the whole vehicle in the target vehicle.

If the scrap grade of the whole vehicle is 4mm, the target vehicle comprises three scrap pressing blocks, the scrap grades of the three scrap pressing blocks are respectively 6mm, 4mm and 2mm, under the condition that the scrap grade of the whole vehicle is 4mm, the scrap pressing block with the scrap grade of 2mm is an unqualified scrap pressing block, and the scrap area ratio of the scrap pressing block in the whole vehicle is the unqualified scrap area ratio.

Meanwhile, the mixing degree of the scrap steel briquettes in the target vehicle can be calculated according to the distribution of the scrap steel grades of each scrap steel briquette in the target vehicle and the impurity proportion of the whole vehicle.

And determining the actual weight of the target vehicle according to the weight difference value between the weight of the whole vehicle scrap steel pressing block contained in the target vehicle and the weight of the target vehicle scrap steel pressing block not contained in the target vehicle.

And finally, inputting at least two of the substandard area ratio of the scrap steel, the mixing degree of the scrap steel pressing blocks in the target vehicle and the weight of the target vehicle into a deduction and deduction model to obtain a deduction and deduction result of the whole scrap steel pressing blocks in the target vehicle.

In the specific implementation, when the deduction and deduction result of the whole vehicle steel scrap pressing block in the target vehicle is calculated, the proportion of special deduction and deduction objects which do not meet the treatment in the target vehicle is also considered, such as an overlong piece and a closed container, because the overlong piece may break a steel furnace, the closed container may cause explosion and the like in the steel melting process, and therefore the special steel materials in the steel scrap pressing block are removed. When calculating the deduction and deduction result of the scrap steel pressing blocks of the whole vehicle, the proportion of special deduction and deduction articles is also considered.

In practical application, counting the unqualified steel scrap area ratio every time below a basic grade according to the judged steel scrap grade of the whole vehicle, obtaining the unqualified steel scrap area ratio kouzhong _ ratio of the whole vehicle by adopting an average value mode for n times of unqualified steel scrap area ratios, estimating the mixing degree mix _ ratio (between 0 and 1) of the steel scrap type of the whole vehicle according to the distribution of different grades of the whole vehicle, obtaining net weight through combining the difference of two weighing times of the vehicle and the proportion ex _ kouzhong _ ratio of special devices such as a super-long member and a closed container for identifying service special deduction, predicting the estimation of the final deduction of the impurities, and obtaining a regression model combined with two or more regression factors of the selected kouzhong _ ratio, kouza _ ratio, mix _ ratio, ex _ kouzhong _ ratio, and a single deduction factor for self-deduction of the whole vehicle, for the regression model, a Machine learning regression algorithm including but not limited to logistic regression, random forest, GDBT (Gradient Boosting Decision Tree), xgboot (eXtreme Gradient Boosting), lightGBM (Light Gradient Boosting Machine, framework for implementing GBDT algorithm) and the like may be selected, or a deep neural network regression model may be selected, and for the training of the regression model, the deduction and deduction of the weight of the historical steel scrap briquette may be used for the regression learning.

Referring to fig. 5, fig. 5 shows a specific processing schematic diagram of an image processing method applied to a scrap steel briquette processing scene according to an embodiment of the present specification.

Specifically, when the scrap steel briquetting is processed by the image processing method, the scrap steel briquetting is interactively realized by an image acquisition module 502, an algorithm module 504 and an application module 506.

The image acquisition module 502 acquires a multi-view image of the scrap steel briquette through multi-angle shooting; and sends the acquired multi-view images of the scrap steel compacts to the algorithm module 504.

An algorithm module 504, which judges the material type of the steel scrap briquette according to the multi-view image of the steel scrap briquette, adjusts the shooting parameters of the cameras at least three views, identifies the unpacking behavior of the unpacking machine on the steel scrap briquette, carries out close-up snapshot on the unpacked steel scrap briquette through the cameras at least three views after the shooting parameters are adjusted under the condition that the unpacking machine identifies the unpacking behavior of the steel scrap briquette, and carries out adulteration identification (for example, identifies impurities and foreign matters in the steel scrap) according to the close-up snapshot image; and when the steel scrap pressing block is determined to have the adulteration situation, sending the close-up snapshot image of the unpacked steel scrap pressing block to the application module 506.

And the application module 506 is used for performing adulteration alarm, manual verification (manual verification whether the adulteration condition exists in the scrap steel pressing block) and result confirmation on a scrap steel processing interface of the terminal.

In practical application, the judgment of the grade of the scrap steel pressing block, the deduction of the weight and the deduction of the impurities and the like are generally realized by a scrap steel grade judging and deduction model at present, but based on the principle of using a generated model, after one-time delivery, the model is not upgraded timely, and the user requirements cannot be met in real time. The specific implementation mode is as follows:

the detection model and the deduction and deduction model form an initial scrap steel grade deduction and penalty model;

correspondingly, after determining the scrap grade and the weight deduction and impurity deduction result of the whole vehicle scrap steel pressing block in the target vehicle according to the scrap steel grade, the impurity proportion and the weight deduction and impurity deduction model of each scrap steel pressing block, the method further comprises the following steps:

displaying the scrap grade of the whole vehicle scrap steel pressing block in the target vehicle and the weight and impurity deduction result to a user through a scrap steel processing interface;

and receiving the modification operation of the user on the scrap steel grade and/or the deduction and deduction result in the scrap steel processing interface, and updating and training the initial scrap steel grade deduction model according to the image of the whole vehicle scrap steel pressing block, the modified scrap steel grade and the deduction and deduction result to obtain a target scrap steel grade deduction model.

Specifically, a user can visually observe the scrap steel grade and the weight and impurity deduction result of the scrap steel pressing block in the target vehicle through a scrap steel processing interface; and under the condition that the scrap steel grade and the weight deduction and impurity deduction results of the scrap steel pressing blocks in the target vehicle are determined to have problems, modifying the scrap steel grade and the weight deduction and impurity deduction results of the scrap steel pressing blocks in the target vehicle at a scrap steel processing interface so as to ensure the accuracy of the scrap steel grade and the weight deduction and impurity deduction results of the scrap steel pressing blocks in the target vehicle. The specific implementation mode is as follows:

the receiving of the modification operation of the user on the scrap steel grade and/or the deduction and impurity deduction result in the scrap steel processing interface comprises:

and receiving modification operation on the scrap grade and/or the deduction and deduction result of the scrap pressing block in the scrap processing interface under the condition that the user confirms that the scrap grade and/or the deduction and deduction result of the scrap pressing block are wrong.

In the embodiment of the specification, the grade judging and penalty deducting system of the scrap steel pressing block can display an editing control for a user on a scrap steel processing interface when the user confirms that any one of the scrap steel grade and/or the weight deduction and impurity deduction result of the scrap steel pressing block has an error, so that the user can modify the scrap steel grade and/or the weight deduction and impurity deduction result of the scrap steel pressing block on the scrap steel processing interface by clicking the editing control, and the accuracy of the scrap steel grade and the weight deduction and impurity deduction result of the scrap steel pressing block in a target vehicle is ensured; and subsequently, updating and training the scrap steel grade deduction and penalty model based on the accurate scrap steel grade of the scrap steel pressing block in the target vehicle and the deduction and heavy deduction result, and further ensuring the accuracy of the scrap steel grade deduction and penalty model.

In practical application, a user can check the scrap grade and the weight-deducting and impurity-deducting result of the scrap steel pressing block in the target vehicle through third-party detection equipment, and under the condition that the scrap steel grade and the weight-deducting and impurity-deducting result of the scrap steel pressing block in the target vehicle detected by the third-party detection equipment are different from the scrap steel grade and the weight-deducting and impurity-deducting result of the scrap steel pressing block in the target vehicle obtained according to the initial scrap steel grade judging and penalty model, the condition that the scrap steel grade and the weight-deducting and impurity-deducting result of the scrap steel pressing block in the target vehicle obtained according to the initial scrap steel grade judging and penalty model are abnormal can be determined. At this time, in order to ensure the accuracy of the scrap grade of the scrap steel briquetting and the deduction and deduction result in the target vehicle, the modification operation of the scrap steel grade of the scrap steel briquetting and/or the deduction and deduction result by the user in the scrap steel processing interface can be received. The specific implementation mode is as follows:

the receiving of the modification operation of the user on the scrap steel grade and/or the deduction and deduction result in the scrap steel processing interface comprises:

receiving the verification result of the user on the scrap grade of the scrap steel pressing block and the deduction and deduction result of the scrap steel pressing block through third-party detection equipment;

and receiving the modification operation of the user on the scrap steel grade and/or the weight deduction and impurity deduction result of the scrap steel pressing block in the scrap steel processing interface under the condition that the verification result is that the scrap steel grade and/or the weight deduction and impurity deduction result of the scrap steel pressing block is wrong.

The third-party detection equipment can be any equipment capable of realizing quality detection on the scrap steel grade and the deduction and impurity deduction results of the scrap steel pressing block in the target vehicle.

In addition, the historical similar operation train number can be directly matched in an image comparison mode, and the scrap steel grade of the scrap steel pressing block in the target vehicle and the weight deduction and impurity deduction result are estimated by referring to the penalty deduction result of the historical train number; and taking the estimated scrap steel grade of the scrap steel pressing block in the target vehicle and the weight deduction and impurity deduction result as a reference, and judging the accuracy of the scrap steel grade of the scrap steel pressing block in the target vehicle and the weight deduction and impurity deduction result obtained according to the initial scrap steel grade judging and penalty model.

And under the condition that the steel scrap grade of the steel scrap pressing block in the target vehicle and the deduction and deduction result obtained according to the initial steel scrap grade deduction and deduction model are abnormal according to the third-party detection equipment, in order to ensure the accuracy of the steel scrap grade of the steel scrap pressing block in the target vehicle and the deduction and deduction result, the steel scrap grade of the steel scrap pressing block in the target vehicle and the deduction and deduction result detected by the third-party detection equipment can be corrected. The specific implementation mode is as follows:

and receiving the modification operation of the user on the scrap steel grade and/or the weight deduction and impurity deduction result of the scrap steel pressing block in the scrap steel processing interface according to the detection result of the third-party detection equipment on the scrap steel pressing block.

And under the condition that modification operation of the user on the scrap steel grade of the scrap steel pressing block in the target vehicle and the deduction and deduction result obtained according to the initial scrap steel grade deduction model is determined, in order to ensure the accuracy of the subsequent scrap steel grade deduction model, the grade deduction and deduction system of the scrap steel pressing block updates and trains the initial scrap steel grade deduction model according to the scrap steel image of the target vehicle, the modified scrap steel grade of the scrap steel pressing block and the deduction and deduction result, and obtains the target scrap steel grade deduction and deduction model. The specific implementation mode is as follows:

according to the image of the whole vehicle scrap steel briquetting, the modified scrap steel grade and the deduction and deduction result, the initial scrap steel grade deduction and deduction model is updated and trained, and a target scrap steel grade deduction and deduction model is obtained, which comprises the following steps:

storing the image of the whole vehicle scrap steel pressing block, the modified scrap steel grade and the deduction and deduction result as sample data to a local database, and adding a problem annotation to the sample data;

in a preset time period, under the condition that the number of the problem comments is larger than or equal to a first number threshold, acquiring all sample data corresponding to the problem comments;

and updating and training the initial scrap steel grade judgment penalty model according to all sample data corresponding to the problem comments to obtain a target scrap steel grade judgment penalty model.

The problem annotation can be understood as the cause of the problem of the sample data, for example, the scrap grade of a part of scrap briquettes in the image of the scrap briquettes cannot be identified, or a part of impurities cannot be identified; moreover, the preset time period and the first number threshold may be set according to practical applications, for example, the preset time period may be set to one day or two days; the first number threshold may be set to 100, 200, etc.

Specifically, the images of all the steel scrap pressing blocks, the modified steel scrap grades and the weight deduction and impurity deduction results of the steel scrap pressing blocks are used as sample data to be stored in a local database, and meanwhile, the images of all the steel scrap pressing blocks of the target vehicle, the modified steel scrap grades and the weight deduction and impurity deduction results of the steel scrap pressing blocks are used as sample data to add problem comments to the sample data; and the system displays the sample data to a user through a scrap steel processing interface and receives the problem comments added to the sample data by the user.

Then, taking a preset time period as one day, and taking a first quantity threshold as 100 as an example; in a preset time period, when the number of the question annotations is greater than or equal to a first number threshold, all sample data corresponding to the question annotations is acquired, which may be understood as that, in one day, when the number of the question annotations is greater than or equal to 100, all sample data corresponding to the question annotations is acquired.

It is understood here that, when the number of sample data to which the same question comment is added is equal to or greater than the first number threshold, the sample data having the same question comment is acquired.

In practical application, after the image of the scrap steel pressing block, the scrap steel grade of the modified scrap steel pressing block and the deduction and deduction result are used as sample data and stored in a local database, the uploaded image of the scrap steel pressing block, the modified scrap steel grade of the scrap steel pressing block and the deduction and deduction result are displayed on a scrap steel processing interface, and a user can carry out sample cleaning and automatic marking on the uploaded image of the scrap steel pressing block, the modified scrap steel grade of the scrap steel pressing block and the deduction and deduction result in the scrap steel processing interface and load final sample data to a training area for model updating training. See in particular fig. 4.

Referring to fig. 6, fig. 6 is a schematic view illustrating a scrap handling interface in an image processing method according to an embodiment of the present disclosure.

In the schematic diagram of the scrap steel processing interface in fig. 6, a model training center is shown, the model training center includes stored images of a plurality of scrap steel compacts, modified scrap steel grades of the scrap steel compacts, and a result of deducting weight and deducting complexity, and the model training center further includes three operation controls of sample cleaning, automatic standardization, and loading to a training area, and a user can perform corresponding operations by clicking the three operation controls.

In specific implementation, if one sample data appears, the initial scrap steel grade judgment penalty model is updated and trained, so that resource waste is caused, and the model training effect is not obvious; if sample data of a plurality of different problems update and train the initial scrap grade judgment penalty model, a large amount of sample data with the same problem can have great influence on the model training process due to different problems of the sample data, and the training effect of a small amount of sample data with the same problem on the model is greatly influenced.

Therefore, in order to solve the problem, the embodiment of the present specification may limit the number of sample data of the model training, for example, when the number of sample numbers with the same problem annotation (i.e. the same exception exists) is greater than or equal to a set number threshold within one day, the update training of the initial steel scrap grade deduction penalty model is triggered to obtain the target steel scrap grade deduction penalty model; the waste of computing resources caused by frequent updating training of the model is avoided; and the effect of sample data of different problems on model training.

After the image of the scrap steel pressing block of the target vehicle, the modified scrap steel grade and the deduction and deduction result of the scrap steel pressing block are stored into the local database as sample data, in order to ensure the availability and accuracy of the sample data, data cleaning and automatic data marking can be carried out on the sample data. The specific implementation mode is as follows:

after the image of the whole vehicle scrap steel briquetting, the modified scrap steel grade and the deduction and impurity deduction result are stored as sample data in a local database, the method further comprises the following steps:

determining images of the scrap steel pressing blocks in all the problem data corresponding to the problem annotations, and the scrap steel grades and the deduction and deduction results of the modified scrap steel pressing blocks;

performing data cleaning on the images of the scrap steel compacts in the same vehicle under the condition that the number of the images of the scrap steel compacts in the same vehicle is larger than a second number threshold;

and taking the cleaned image of each scrap steel pressing block in the same vehicle as a training sample, and taking the scrap steel grade and the deduction and impurity deduction result of the scrap steel pressing block modified by the same vehicle as a training label of each training sample.

Specifically, determining an image of a scrap steel pressing block of the target vehicle and a problem annotation corresponding to sample data formed by the modified scrap steel grade of the scrap steel pressing block and the deduction and deduction result; and acquiring all sample data which are the same as the problem comments, namely images of the steel scrap briquettes of all vehicles with the same problem, and the steel scrap grades and the weight deduction and impurity deduction results of the modified steel scrap briquettes.

Data cleaning is carried out on the images of the scrap steel briquettes, such as incomplete, exposed and repeated images of the scrap steel briquettes are deleted; and automatically marking the waste steel image cleaned according to the data of each vehicle, the corresponding modified waste steel grade of the waste steel pressing block and the deduction and deduction result.

In practical application, because the number of the scrap steel images of some vehicles is small and the number of the scrap steel briquetting images of some vehicles is large, if the number of the scrap steel briquetting images is small, data cleaning is performed, all the scrap steel briquetting images of the vehicles can be deleted, and sample data cannot be formed to participate in model training; in order to avoid this situation, if the number of images of scrap pieces of the same vehicle is greater than or equal to a second number threshold (for example, 10 sheets), the data of the images of scrap pieces of the vehicle may be cleaned. However, it is not excluded that, when the image of the scrap steel briquette of a certain vehicle, and the corresponding scrap steel grade and weight deduction result of the modified scrap steel briquette have little influence on the model training, the data cleaning may be performed, and the determination may be specifically performed according to the actual requirements.

And after data cleaning, taking the image of each scrap steel pressing block in the same vehicle as a training sample, taking the modified scrap steel grade and the deduction and deduction result of the scrap steel pressing block corresponding to the image of the scrap steel pressing block in the same vehicle as a training label corresponding to each training sample, and updating and training the initial scrap steel grade deduction model by taking the training samples and the training labels of all vehicles as sample data to obtain a target scrap steel grade deduction model. The specific implementation mode is as follows:

the updating and training of the initial steel scrap grade judgment penalty model according to all sample data corresponding to the problem comments to obtain a target steel scrap grade judgment penalty model comprises the following steps:

and updating and training the initial scrap steel grade deduction penalty model according to training samples and training labels in all sample data corresponding to the problem comments to obtain a target scrap steel grade deduction penalty model.

The above is a detailed description of the update training of the model on the terminal, and in practical application, in order to reduce the computational stress of the terminal, the update training process of the model may be placed in the cloud for update training. The specific implementation mode is as follows:

after obtaining all sample data corresponding to the question annotation, the method further includes:

sending all sample data corresponding to the problem annotation to a server;

and receiving all sample data which are sent by the server and correspond to the problem annotations, and updating and training the initial steel scrap judging penalty model to obtain a target steel scrap judging penalty model.

The server can be understood as a common server or a cloud, and a user can send all sample data corresponding to the problem annotation to the server by logging in a cloud account; after receiving the sample data, the server side can perform data cleaning and automatic marking on the sample data in the manner of the embodiment to form final sample data, and after performing updating training on the model through the sample data, the server side returns to the grade judgment penalty system of the scrap steel pressing block at the terminal.

During specific implementation, in order to further reduce the calculation pressure of the terminal, after the image of the scrap steel pressing block of each target vehicle, the modified scrap steel grade of the scrap steel pressing block and the weight deduction and impurity deduction result are obtained, the image, the modified scrap steel grade of the scrap steel pressing block and the impurity deduction result are sent to the server, the server adds problem annotation, data cleaning, automatic marking and model updating training for the image, and finally the model after updating training is returned to the terminal for deployment. The specific implementation mode is as follows:

the updating training of the initial scrap steel grade deduction penalty model according to the image of the whole vehicle scrap steel pressing block, the modified scrap steel grade and deduction penalty result to obtain a target scrap steel grade deduction penalty model comprises the following steps:

sending the image of the whole vehicle scrap steel pressing block, the modified scrap steel grade and the deduction and impurity deduction result as sample data to a server;

and receiving a target scrap steel grade deduction penalty model which is sent by the server and is obtained by carrying out updating training on the initial scrap steel grade deduction penalty model according to the sample data.

In practical application, in order to ensure the performance of the model in the using process, the model is subjected to performance evaluation after being updated, and the model is deployed at the terminal only when the performance evaluation meets the requirements. The specific implementation mode is as follows:

after the target scrap steel grade deduction penalty model is obtained, the method further comprises the following steps:

and under the condition that the evaluation result of the target scrap steel grade deduction model meets the evaluation condition, replacing the initial scrap steel grade deduction model with the target scrap steel grade deduction model.

The evaluation condition may be set according to actual application, for example, the evaluation condition may set the evaluation result to be greater than 75 points.

Specifically, under the condition that the evaluation result of the target scrap steel grade deduction model meets the evaluation condition, the target scrap steel grade deduction model is replaced with the initial scrap steel grade deduction model.

Referring to fig. 7, fig. 7 shows a schematic diagram of an evaluation result of a target scrap box grade deduction model in a scrap box grade deduction penalty model updating method of self-iterative learning provided by an embodiment of the present specification.

Taking the evaluation condition as an example that the evaluation result is more than 70 points, according to the display content of the model training center page of the scrap steel processing interface in the graph 7, the evaluation result of the target scrap steel grade deduction model is 73.88; at this time, it may be determined to replace the initial scrap grade deduction model with the target scrap grade deduction model.

In the embodiment of the specification, when model updating training is performed, sample data is actually divided into training data and test data, the model is updated and trained by the training data, after the updated model is obtained, the updated model is evaluated according to the test data, and under the condition that an evaluation result meets a preset evaluation condition, the performance of a target scrap steel grade deduction penalty model obtained after updating can be determined to be better, and at the moment, the target scrap steel grade deduction penalty model can replace an initial scrap steel grade deduction penalty model, that is, the target scrap steel grade deduction penalty model is deployed in a grade deduction penalty system of a scrap steel pressing block of a terminal for grade deduction penalty.

In the method for updating the scrap steel grade judgment penalty model through self-iterative learning provided by the embodiment of the specification, in the operation process of the initial scrap steel grade judgment penalty model, if a user finds that a model result is abnormal, the model result can be modified through a scrap steel processing interface, an abnormal scrap steel image and the modified result are taken as data samples to flow back, the initial scrap steel grade judgment penalty model is updated and trained through the data samples, and the target scrap steel grade judgment penalty model is obtained, so that the model can be updated in time in an iterative manner, and the requirement of the user can be met in real time.

In addition, the updating of the scrap steel grade deduction penalty model of self-iterative learning in the image processing method provided by the embodiment of the description can be applied to a common scrap steel deduction penalty scene.

The following describes further updating of the scrap steel grade deduction model of self-iterative learning in the image processing method provided in this specification, with reference to fig. 8, by taking an application of the scrap steel grade deduction model of self-iterative learning in a general scrap steel grade deduction and penalty scene as an example. Fig. 8 shows a flow chart of a processing procedure of updating a scrap judgment penalty model through self-iterative learning in an image processing method provided in an embodiment of the present specification, and specifically includes the following steps.

Fig. 8 includes three parts: the reasoning system of the part a, the project system of the part b and the training system of the part c; the system comprises an inference system, a deduction system and a deduction system, wherein the inference system is used for judging the grade of the scrap steel and determining a deduction and deduction result through a scrap steel grade deduction and penalty algorithm; the project system is used for displaying the grade of the scrap steel calculated by the reasoning system and the deduction and deduction result, and carrying out settlement according to the grade of the scrap steel and the deduction and deduction result under the condition that the grade of the scrap steel and the deduction and deduction result are accurate; and the training system is used for receiving the modification result of the grade of the scrap steel and the deduction and deduction result of the scrap steel and the image of the scrap steel by the settlement system under the condition that the grade of the scrap steel and the deduction and deduction result of the scrap steel are judged to be inaccurate by the user, and updating and training the grade deduction and deduction algorithm of the scrap steel in the inference system by taking the modification result as sample data.

Step 802: and the truck is parked and begins to unload.

In particular, the truck is parked to begin dumping, which is to be understood as meaning that a truck loaded with scrap is parked in the dumping zone and begins to unload the loaded scrap to the dumping zone.

Step 804: and (5) unloading and shooting.

Specifically, the shooting of unloading can be understood as that the truck photographs the poured scrap steel through a camera fixedly arranged at the periphery of the truck or a camera (such as an unmanned aerial vehicle) movably arranged at the periphery of the truck in the process of unloading the scrap steel, so as to obtain a first layer image and a second layer image of the scrap steel.

Step 806: and (4) algorithm identification.

Specifically, the algorithm identification may be understood as inputting the shot first layer image and the shot second layer image of the scrap steel into a scrap steel grade judgment penalty model trained in advance for identification, and identifying the grade, the impurity proportion and the like of the scrap steel in each layer image, so as to obtain a first layer result, a second layer result, an nth layer result and the like.

Step 808: and (5) carrying out level weighing and impurity deduction.

Specifically, the scrap steel grade and the weight and impurity deduction results of the whole truck are obtained according to the first layer result and the second layer result.

Step 810: and (5) displaying the settlement system.

Specifically, the settlement system display can be understood as displaying the scrap level and the withholding and impurity-deducting results of the whole truck to a user through a scrap processing interface in the settlement system. If the user determines that the scrap steel grade of the whole vehicle and the deduction and impurity deduction result are correct, carrying out scrap steel settlement in a settlement system; and if the user determines that the scrap steel grade of the whole vehicle and the deduction and deduction result are wrong, performing manual correction on the scrap steel processing interface.

Step 812: and correcting the sample reflux library.

Specifically, the sample reflux library is corrected, that is, the scrap image of the truck, the corrected scrap grade and the deduction and deduction result of the finished vehicle are taken as sample data to be refluxed to the local sample library or the cloud sample library, and data labeling is performed in the sample library.

Step 814: a model training framework is determined.

Specifically, determining a model training framework can be understood as performing self-help algorithm optimization by using an embedded model training framework through the sample data.

Step 816: and putting the model into a model generation library.

Specifically, the model is placed in a model generation library, which can be understood as that the model obtained by each iterative training is placed in the model generation library.

Step 818: and (6) evaluating the model.

Specifically, the model evaluation can be understood as performing comprehensive evaluation on multiple training results through a model training framework to give an optimal model.

Step 820: and (6) deploying the model.

Specifically, the optimized optimal model deployment is issued to the existing operating environment through a one-key derivation function of the system, so that version upgrade/function addition of the model is realized.

The method for updating the scrap steel grade judgment penalty model through self-iterative learning provided by the embodiment of the specification realizes self-closed loop upgrading processes of production, optimization, deployment, iterative self-learning of a system algorithm, automatic completion of problem data collection, marking, model updating training, optimal model deployment, adaptation and the like; on the production side, the method uses a deep learning image processing technology to realize automatic identification of the grade of the steel scrap on the vehicle and detection and identification of various impurities, and then combines industry history punishment rule data to carry out statistical analysis of big data, thereby realizing automatic calculation of the grade of the steel scrap of the whole vehicle and punishment; in the aspect of self-learning, in the operation process of the system (the whole self-iterative learning system), if a user finds that the output result of the reasoning system is different from the accurate grading result greatly, the problem data can be confirmed, then the output result of the reasoning system is revised and recorded through the result re-judging function of the interface, the upgrading function of the reasoning system triggers the backflow of corresponding data samples and uploads the backflow to a sample library for data marking, model training, testing and redeployment, and finally the self-optimization upgrading closed loop of the algorithm model is formed.

Corresponding to the above method embodiment, the present specification further provides an image processing apparatus embodiment, and fig. 9 shows a schematic structural diagram of an image processing apparatus provided in an embodiment of the present specification. As shown in fig. 9, the apparatus is applied to scrap steel briquetting treatment, comprising:

an image acquisition module 902 configured to acquire at least one set of unpacking images of each scrap compact, wherein each scrap compact is acquired from a target vehicle and the at least one set of unpacking images of each scrap compact is acquired from at least three different perspectives;

a model processing module 904, configured to input the at least one set of unpacking images of each steel scrap pressing block into a detection model, and obtain different grades of steel scrap proportions and impurity proportions in the at least one set of unpacking images of each steel scrap pressing block, wherein the detection model is a machine learning model;

a determining module 906 configured to determine a scrap grade and an impurity proportion of each scrap briquette according to the scrap proportions and impurity proportions of different grades in at least one set of unpacking images of each scrap briquette;

a result obtaining module 908 configured to determine a scrap grade and a deduction and deduction result of the whole vehicle scrap steel compacts in the target vehicle according to the scrap grade, the impurity proportion and the deduction and deduction model of each scrap steel compact.

Optionally, the image acquisition module 902 is further configured to:

acquiring panoramic images of each scrap steel pressing block at different visual angles through at least three pieces of shooting equipment at different visual angles;

determining the unpacking position of each steel scrap pressing block according to the panoramic image of each steel scrap pressing block at different viewing angles;

Optionally, the image acquisition module 902 is further configured to:

Optionally, the image obtaining module 902 is further configured to:

Optionally, the image acquisition module 902 is further configured to:

and under the condition that power-off instructions sent by the unpacking equipment for unpacking each scrap steel pressing block are received, acquiring at least one group of unpacking images of each scrap steel pressing block at different visual angles through the adjusted at least three shooting equipment at different visual angles.

Optionally, the model processing module 904 is further configured to:

Optionally, the determining module 906 is further configured to:

and obtaining the impurity proportion of each scrap steel pressing block in a weighting and averaging mode according to the impurity proportion of each group of unpacking images.

Optionally, the result obtaining module 908 is further configured to:

determining the whole vehicle scrap grade and the whole vehicle impurity ratio of the whole vehicle scrap pressing blocks in the target vehicle according to the scrap grade and the impurity ratio of each scrap pressing block;

Optionally, the result obtaining module 908 is further configured to:

determining the mixing degree of the scrap steel pressing blocks in the target vehicle according to the distribution of the grade of the scrap steel of each scrap steel pressing block in the target vehicle and the impurity proportion of the whole vehicle;

determining the weight of a target vehicle containing finished vehicle scrap steel pressing blocks and the weight of a target vehicle not containing finished vehicle scrap steel pressing blocks, and determining the difference value of the weight of the target vehicle containing finished vehicle scrap steel pressing blocks and the weight of the target vehicle not containing finished vehicle scrap steel pressing blocks as the weight of the target vehicle;

and inputting at least two of the substandard area ratio of the waste steel, the mixing degree of the waste steel pressing blocks in the target vehicle and the weight of the target vehicle into a deduction and deduction model to obtain a deduction and deduction result of the whole waste steel pressing blocks in the target vehicle.

Optionally, the detection model and the deduction and deduction model form an initial scrap steel grade deduction and penalty model;

the device, still include:

a model update module configured to:

Optionally, the model update module is further configured to:

storing the image of the whole vehicle scrap steel pressing block, the modified scrap steel grade and the deduction and impurity deduction result as sample data to a local database, and adding a problem annotation to the sample data;

and updating and training the initial scrap steel grade judgment and penalty model according to all sample data corresponding to the problem comments to obtain a target scrap steel grade judgment and penalty model.

Optionally, the model update module is further configured to:

The image processing device provided by the embodiment of the specification adopts a multi-angle imaging technology, acquires unpacking images of steel scrap pressing blocks from at least three different visual angles, performs multi-angle analysis on the unpacking process of the steel scrap packing materials and the pressing blocks by combining a method of integral and local analysis, automatically identifies the grades of the steel scrap packing materials and the pressing blocks by a detection model, and realizes automatic calculation of the weight and impurity of the steel scrap packing materials and the pressing blocks by combining a weight and impurity deduction model; meanwhile, the detection and identification of impurities in the unpacking process of the steel scrap packing material and the briquetting can be realized.

The above is a schematic configuration of an image processing apparatus of the present embodiment. It should be noted that the technical solution of the image processing apparatus belongs to the same concept as the technical solution of the image processing method, and details that are not described in detail in the technical solution of the image processing apparatus can be referred to the description of the technical solution of the image processing method.

Another embodiment of the present specification provides an image processing interaction method, which is applied to an image processing interaction system, where the system includes at least three capturing devices with different viewing angles, an unpacking device, and an image processing apparatus, where the method includes:

inputting the at least one group of unpacking images of each scrap steel pressing block into a detection model to obtain scrap steel areas with different grades and impurity areas in the at least one group of unpacking images of each scrap steel pressing block, wherein the impurity areas are other areas except the scrap steel areas,

determining the scrap grade and the impurity proportion of each scrap steel briquetting according to the scrap steel areas and the impurity areas with different grades in at least one group of unpacking images of each scrap steel briquetting,

FIG. 10 illustrates a block diagram of a computing device 1000 provided in accordance with one embodiment of the present description. The components of the computing device 1000 include, but are not limited to, memory 1010 and a processor 1020. The processor 1020 is coupled to the memory 1010 via a bus 1030 and the database 1050 is used to store data.

Computing device 1000 also includes access device 1040, access device 1040 enabling computing device 1000 to communicate via one or more networks 1060. Examples of such networks include the Public Switched Telephone Network (PSTN), a Local Area Network (LAN), a Wide Area Network (WAN), a Personal Area Network (PAN), or a combination of communication networks such as the internet. Access device 1040 may include one or more of any type of network interface, e.g., a Network Interface Card (NIC), wired or wireless, such as an IEEE802.11 Wireless Local Area Network (WLAN) wireless interface, a worldwide interoperability for microwave access (Wi-MAX) interface, an ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a bluetooth interface, a Near Field Communication (NFC) interface, and so forth.

In one embodiment of the present description, the above-described components of computing device 1000 and other components not shown in FIG. 10 may also be connected to each other, such as by a bus. It should be understood that the block diagram of the computing device architecture shown in FIG. 10 is for purposes of example only and is not limiting as to the scope of the present description. Those skilled in the art may add or replace other components as desired.

Computing device 1000 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., tablet, personal digital assistant, laptop, notebook, netbook, etc.), a mobile phone (e.g., smartphone), a wearable computing device (e.g., smartwatch, smartglasses, etc.), or other type of mobile device, or a stationary computing device such as a desktop computer or PC. Computing device 1000 may also be a mobile or stationary server.

Wherein the processor 1020 is configured to execute computer-executable instructions that, when executed by the processor, implement the steps of the image processing method described above.

The above is an illustrative scheme of a computing device of the present embodiment. It should be noted that the technical solution of the computing device and the technical solution of the image processing method belong to the same concept, and details that are not described in detail in the technical solution of the computing device can be referred to the description of the technical solution of the image processing method.

An embodiment of the present specification further provides a computer-readable storage medium storing computer-executable instructions, which when executed by a processor implement the steps of the image processing method described above.

The above is an illustrative scheme of a computer-readable storage medium of the present embodiment. It should be noted that the technical solution of the storage medium belongs to the same concept as the technical solution of the image processing method, and details that are not described in detail in the technical solution of the storage medium can be referred to the description of the technical solution of the image processing method.

An embodiment of the present specification further provides a computer program, wherein when the computer program is executed in a computer, the computer is caused to execute the steps of the image processing method.

The above is an illustrative scheme of a computer program of the present embodiment. It should be noted that the technical solution of the computer program belongs to the same concept as the technical solution of the image processing method, and for details that are not described in detail in the technical solution of the computer program, reference may be made to the description of the technical solution of the image processing method.

The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

The computer instructions comprise computer program code which may be in the form of source code, object code, an executable file or some intermediate form, or the like. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media does not include electrical carrier signals and telecommunications signals as is required by legislation and patent practice.

It should be noted that, for the sake of simplicity, the foregoing method embodiments are described as a series of acts, but those skilled in the art should understand that the present embodiment is not limited by the described acts, because some steps may be performed in other sequences or simultaneously according to the present embodiment. Furthermore, those skilled in the art will appreciate that the embodiments described in this specification are presently preferred and that no acts or modules are required in the implementations of the disclosure.

In the foregoing embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to the related descriptions of other embodiments.

The preferred embodiments of the present specification disclosed above are intended only to aid in the description of the specification. Alternative embodiments are not exhaustive and do not limit the invention to the precise embodiments described. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the embodiments and the practical application, to thereby enable others skilled in the art to best understand and utilize the embodiments. The specification is limited only by the claims and their full scope and equivalents.

Claims

1. An image processing method is applied to scrap steel briquetting processing and comprises the following steps:

adjusting shooting parameters of shooting equipment with at least three different visual angles according to the unpacking position of each scrap steel pressing block, and acquiring at least one group of unpacking images of each scrap steel pressing block through the adjusted shooting equipment with at least three different visual angles, wherein each scrap steel pressing block is acquired from a target vehicle, and the unpacking position is an initial placement position of each scrap steel pressing block on the unpacking equipment;

inputting the at least one group of unpacking images of each steel scrap briquetting into a detection model to obtain different grades of steel scrap proportions and impurity proportions in the at least one group of unpacking images of each steel scrap briquetting, wherein the detection model is a machine learning model, the inputting the at least one group of unpacking images of each steel scrap briquetting into the detection model to obtain different grades of steel scrap proportions and impurity proportions in the at least one group of unpacking images of each steel scrap briquetting comprises inputting each unpacking image of each group of unpacking images of each steel scrap briquetting into the detection model to obtain different grades of steel scrap regions and impurity regions in each unpacking image, and determining different grades of steel scrap proportions and impurity proportions in each unpacking image according to the different grades of steel scrap regions and impurity regions in each unpacking image, determining the steel scrap proportion and the impurity proportion of different grades in at least one group of unpacking images and each group of unpacking images of each steel scrap briquetting according to the steel scrap proportion and the impurity proportion of different grades in each unpacking image;

determining the scrap grade and the impurity proportion of each scrap briquetting according to the scrap proportions and the impurity proportions of different grades in at least one group of unpacking images of each scrap briquetting;

and determining the scrap grade and the deduction and deduction results of the whole scrap steel pressing blocks in the target vehicle according to the scrap steel grade and the impurity proportion of each scrap steel pressing block and the deduction and deduction model.

2. The image processing method of claim 1, wherein said obtaining at least one set of unpacking images of each scrap steel compact comprises:

3. The image processing method according to claim 2, wherein the adjusting of the shooting parameters of the shooting equipment with at least three different visual angles according to the unpacking position of each scrap steel pressing block comprises the following steps:

determining a shooting focus position and a shooting magnification factor according to the unpacking position of each scrap steel pressing block;

4. The image processing method according to claim 2, wherein the step of acquiring the panoramic image of each scrap steel briquette at different viewing angles through at least three shooting devices at different viewing angles comprises the following steps:

5. The image processing method according to claim 2, wherein the acquiring at least one group of unpacking images of each steel scrap pressing block at different viewing angles through the adjusted at least three shooting devices at different viewing angles comprises:

6. The image processing method according to claim 1, wherein the determining of the scrap grade and the impurity proportion of each scrap briquette according to the scrap proportion and the impurity proportion of different grades in at least one group of unpacking images of each scrap briquette comprises:

7. The image processing method according to claim 1, wherein the determining the scrap grade and the weight deduction and impurity deduction result of the whole scrap steel briquettes in the target vehicle according to the scrap steel grade, the impurity proportion and the weight deduction and impurity deduction model of each scrap steel briquette comprises:

and determining a deduction and deduction result of the whole vehicle scrap steel pressing block in the target vehicle according to the whole vehicle scrap steel grade, the whole vehicle impurity proportion and a deduction and deduction model.

8. The image processing method according to claim 7, wherein the determining of the deduction and deduction result of the whole vehicle scrap steel pressing block in the target vehicle according to the whole vehicle scrap steel grade, the whole vehicle impurity proportion and the deduction and deduction model comprises:

according to the whole vehicle scrap grade, determining the area proportion of unqualified scrap steel which does not meet the whole vehicle scrap grade in a whole vehicle scrap steel pressing block in the target vehicle;

9. The image processing method according to claim 1, wherein the detection model and the deduction and deduction model form an initial steel scrap grade deduction and deduction model;

10. The image processing method according to claim 9, wherein the updating and training of the initial scrap steel grading and penalty deducting model according to the image of the whole vehicle scrap steel briquetting and the modified scrap steel grade and penalty deduction result to obtain a target scrap steel grading and penalty deduction model comprises:

11. The image processing method according to claim 10, further comprising, after storing the image of the entire car scrap briquette, the modified scrap grade and the deduction and deduction result as sample data in a local database:

determining images of the scrap steel pressing blocks in all the problem data corresponding to the problem comments, and the scrap steel grade and the deduction and impurity deduction results of the modified scrap steel pressing blocks;

12. The image processing method according to claim 9, wherein the updating and training of the initial scrap steel grading and penalty deducting model according to the image of the whole vehicle scrap steel briquetting and the modified scrap steel grade and penalty deduction result to obtain a target scrap steel grading and penalty deduction model comprises:

13. An image processing interaction method is applied to an image processing interaction system, the system comprises at least three shooting devices with different view angles, an unpacking device and an image processing device, wherein the method comprises the following steps:

inputting the at least one group of unpacking images of each steel scrap briquetting into a detection model to obtain different grades of steel scrap proportions and impurity proportions in the at least one group of unpacking images of each steel scrap briquetting, wherein the detection model is a machine learning model, the inputting the at least one group of unpacking images of each steel scrap briquetting into the detection model to obtain different grades of steel scrap proportions and impurity proportions in the at least one group of unpacking images of each steel scrap briquetting comprises inputting each unpacking image of each group of unpacking images of each steel scrap briquetting into the detection model to obtain different grades of steel scrap regions and impurity regions in each unpacking image, and determining different grades of steel scrap proportions and impurity proportions in each unpacking image according to the different grades of steel scrap regions and impurity regions in each unpacking image, determining the steel scrap proportion and the impurity proportion of different grades in at least one group of unpacking images and each group of unpacking images of each steel scrap pressing block according to the steel scrap proportion and the impurity proportion of different grades in each unpacking image;